How to Read the Atlas
The Microbiome Signal Atlas is a structured evidence map. It is designed to help you judge whether a microbiome signal looks substantial, human-relevant, disease-linked, or possibly overhyped. It is not a clinical decision tool, not medical advice, and not a claim that every paper in a field has been perfectly resolved.
What this is
This atlas organizes published evidence for microbial signals across disease contexts. Each signal-disease pair is summarized using evidence categories, count structure, and a first-pass judgment layer.
The point is not just to find papers. The point is to help you decide whether a signal looks credible, human-grounded, and worth taking seriously.
What the evidence categories mean
Human or disease-relevant primary papers that look directly important to the signal in the specified disease context.
Primary papers that are relevant but less central. These may be broader association studies or papers where the signal matters but is not the core focus.
Mouse, model-system, mechanistic, organoid, or other non-human evidence that may support biological plausibility but should not be confused with strong human validation.
Reviews and meta-analyses that show field attention and synthesis, but do not replace direct human primary evidence.
What the scores mean
Signal credibility score
A 0 to 100 summary score based on the current evidence mix. Higher scores generally reflect stronger central evidence, more human support, and less dependence on review-only or preclinical-heavy literature.
This is a judgment aid, not a claim of truth.
Human translation score
A 1 to 5 estimate of how human-grounded the signal looks. Higher scores mean the evidence leans more toward human primary data rather than mostly preclinical literature.
A high score does not prove clinical utility.
Reproducibility score
A 1 to 5 estimate of how repeatedly the signal appears across the literature. More central papers and strong review density tend to push this upward.
Repeated appearance is not the same as mechanistic proof.
Novelty score
A 1 to 5 estimate of how under-saturated the signal is. Lower scores usually mean the signal is already widely discussed. Higher scores suggest a less crowded or less overworked signal.
Novel does not automatically mean important.
Actionability score
A 1 to 5 estimate of whether the signal looks potentially useful for biomarker thinking, translational prioritization, or intervention framing.
This is not a therapeutic recommendation.
Claim inflation risk
A qualitative estimate of whether the literature mix looks vulnerable to overclaiming. Signals can have lots of papers and still carry high inflation risk if review density is high and strong human central evidence is limited.
This is meant to slow you down, not flatter your favorite signal.
How to interpret the signal landscape map
Bubble size reflects the signal credibility score. Bigger bubbles mean the atlas currently rates that signal-disease pair as more credible overall.
Color reflects claim inflation risk. Green means lower inflation risk, amber means moderate risk, and red means higher risk of overclaiming relative to the evidence mix.
Border thickness reflects the human translation score. Thicker borders mean the signal looks more human-grounded.
Use the map to see where signals cluster, which diseases appear crowded, and where a signal looks stronger, weaker, or more hype-prone across contexts.
What this does not mean
- This is not medical advice.
- This is not a clinical diagnostic system.
- This does not prove causality.
- This does not guarantee strain-level resolution.
- This does not mean a higher score equals therapeutic validity.
- This is not a substitute for reading key papers in detail.
Frequently asked questions
Are these scores manually curated or automated?
Right now the scores are generated from structured evidence patterns and simple judgment logic. They are designed to be useful and skeptical, not magical. Some dimensions are easier to estimate automatically than others.
Why can a signal have lots of papers and still high inflation risk?
Because volume is not the same thing as quality. A signal can be surrounded by review articles, preclinical enthusiasm, or generic dysbiosis literature without having deep central human evidence.
Why does the same signal appear in multiple diseases?
Many microbial signals are not disease-specific. Some reflect broad inflammatory states, altered ecology, barrier dysfunction, antibiotics, or generic metabolic disturbance. Cross-disease appearance is part of what this atlas is meant to reveal.
Why are some familiar signals missing?
The atlas is still expanding. Coverage depends on what has been ingested into the current build and how well the signal is represented in the structured retrieval pipeline.
Are fungi, pathways, ecology markers, and taxa all treated the same way?
No. They can all be represented in the atlas, but the evidence patterns are different. A taxonomy-centered literature often behaves differently from functional or ecological signals, and fungal evidence can be especially uneven.
How often is the atlas updated?
This depends on curation and rebuild cycles. The current version should be treated as a living beta rather than a frozen reference work.
What should I do if a result looks wrong or incomplete?
Treat that as feedback, not apocalypse. The atlas is meant to support judgment, not replace it. Obvious misses, noisy matches, and edge cases are expected in a living system.